The Advanced Predictive System (APS) processes and manages the data coming from the DAS through the wireless network (WCN) to predict and optimise the injection moulding process. The predictions and recommendations produced by the APS are stored in a centralised server called the content management system (CMS) for future analysis and sent to the mobile app of the location based content delivery via a standard wireless network.
The APS system applies novel and advanced artificial intelligence and machine learning techniques which create the robust and accurate prediction and optimisation of the injection moulding processes compared to the traditional statistical process control (SPC) currently available in the market.
The PREVIEW APS system has two subsystems-
1. Setup Predictive Systems (SPS): The SPS subsystem is responsible for providing recommendations to tune key parameters of the production process for a particular mould/machine configuration. This feature is capable of providing the following: recommendations, predictions and alarms; representative examples of these are provided below:
• SPS Recommendations: Example – “Increase holding pressure 11%”.
• SPS Alarm: Example – “Possible bad pieces in less than 10 cycles, Production is deviating towards holding pressure: 800”.
2. Production Control System (PCS): This subsystem monitors the production process and offers full trace-ability of the production, predicts parts quality, provides feedback to the process with identification of defect types and triggers early and predictive alarms to correct the production process. Some examples of part quality and defect prediction, PCS prediction and statistics, and alarms, are as follows:
• Part Quality Prediction and Defect Type: Example-“Good parts”, “Bad parts-Shrinkage”, “Bad parts-Flash”.
• PCS Alarm: Example – “Producing bad parts, possible bad pieces in less than 10 cycles”.
Both of the subsystems work based on the PREVIEW system’s Mould-DNA concept. In Mould-DNA a machine learning model is generated for a specific mould-machine configuration from trials (training phase) carried out previously by the operator which contain data of the mould cavity sensors and machine configurations, mould cavity pressure and temperature, machine parameters and the quality of produced parts.
The PCS subsystem analyses the mould cavity pressure and temperature curves along with the machine parameters and compares them with the Mould-DNA model which creates real-time prediction of the part quality and launches production alerts and early alarms for the deviated production. The SPS module analyses the cavity data against the optimum machine parameters to identify process deviations.
The APS system has the full compatibility and seamless integration with the INLFATE protocol designed for the PREVIEW wireless network and also with the central server. The most attractive functionalities of the APS system are as follows-
1. Parallel processing of multi moulds/machines in the production floor.
2. Data fusion for combining the information of different sensors.
3. Automatic identification of data incoherency and raising of alarms.
4. Less samples required for the training of the system.
5. Part quality and types of defect prediction at more than 95% of accuracy.
6. Warnings and preventive alarms in a few injection cycles for production quality deviation.
7. Machine configuration identification up-to 80% of accuracy.
8. Machine tuning recommendations in an iterative process.
9. Storage of the processed data for creating an historical database.
10. Advanced production control and set-up recommendations statistics display.
For further details feel free to consult:
 Changyu Shen, L. W. Optimization of injection molding process parameters using
combination of artificial neural network and genetic algorithm method. Journal of Materials
Processing Technology, (pp. 412–418).
. B.Kayis, S. K. (2007). Set-Up Reduction in Injection Molding Process – A Case Study in
Packaging Industry. 4th International Conference and Exhibition on Design and Production of
MACHINES and DIES/MOLDS.
. Manjunath Patel, P. K. (2013). A Review on Application of Artificial Neural Networks for
Injection Moulding and Casting processes. International Journal of Advances in Engineering
. P.K. Bharti, M. I. (2012). Recent methods for optimization of plastic injection molding
process –a retrospective and literature review. Popular Plastics & Packaging, (p. 26).
PREVIEW Whitepaper (July 2017)
Read the whitepaper on PREVIEW and find out about how the system works, its four sub-systems and the testimonials of the benefits of PREVIEW in action!
Advanced Predictive System – Interview
Francesc Bonada, R&D Project Manager at Eurecat Centre Tecnològic de Catalunya tells us about the PREVIEW Advanced Predictive System and how it forms the overall part of the PREVIEW concept.